RAP

What is “RAP”?

🎤

RAP is the gold standard of analysis.
I didn’t know what RAP is for a long time and always had it on my list to find out more about. When I finally did I was so glad that I spent the time to do so.
Also figured it is a lot less scary than initally thought and a lot of us are doing RAP already.
What is the simplest definition of RAP?
Well, its creating a piece of analysis that is fully reproducible.
i.e., With the same data and the same analysis, every time you run this you get the same results.
No matter the software, version, operating system, location or time.
But, this view of RAP is more tailored to academia and publication of results/studies (Gov publications).
RAP is broader than this and can be interpreted differently (to be made more appropriate for the work we do in our unit).

Why a pipeline?

So… How do we deal with this?

Become a…

You’re likely already implementing RAP

(or a variation of it)

The Reproducibilty Iceberg

An expanding brain meme styled iceberg showing the development of RAP.

Why should you RAP?

What happens if you don’t?

Messy files and folders

Confusing names

Lost knowledge

No papertrail, rollback or accountability

Bugs and errors

Duplicated work and low efficiency

Why should you RAP?

An expanding brain meme styled iceberg showing the development of RAP. — When they find out you RAP

It will make your life

easier

Your work will be of

higher quality

Work should be more

enjoyable

and even

fun!

May mean slightly more work in the short term, but in the long term it will save a lot of time. Both from not having to repeat work, to less errors and it being easier to pass on to other colleagues.
Higher quality work is always better. Your work will be robust and easy to share, important when taking it to that next step (e.g., another project or passing over to other teams). If there are ever any audits of your work (which is more likely that you think) then this will put your mind at ease.
Enjoyable by the fact you can do new things and no longer be bogged down by the repetitive tasks. Collaboration should increase, and working together is often overlooked in data analytics.
Fun can be had by trying out advanced coding, creating products that look and feel cool and are using state-of-the-art technology. (For example, creating your own package, automating workflows and building websites.)

Simple steps to achieve Reproducibility

Tidy files and folders 🧹

Effective naming 👨‍👩‍👦‍👦

Document like you ctrl S ✍️

Version control 🛠️

Software stability 🤸

Share, publish/open code, results… 🌎

Where to start?

Implement the 6 steps 👟

Checkout the DfE’s RAP strategy 🎲

Ask me for help! 🙋

References

Seibold. H., (10th Oct 2023). RAP webinar October 2023. Health Data Research UK. Available YouTubeat: https://www.youtube.com/watch?v=C27ubS0AioA. (Accessed: 20 December 2023)

Rodrigues. B., (October 3, 2023). Building reproducible analytical pipelines with R. Available at: https://raps-with-r.dev/ (Accessed: 20 December 2023)